The absence of the scaling factors ωl means that, compared with mash, their model is less well suited to capturing effects that have similar patterns (relative sizes across conditions) but vary in magnitude. By applying it to only the largest effects, we seek to sidestep this issue.
